Dear Editor

We thank Professor McLeod for his interest in our paper, “Vitrectomy with Silicone Oil Infusion in Severe Diabetic Retinopathy”[1] and value the opportunity to address some of the issues he has raised.[2]

Professor McLeod notes that surgical success can be judged only if the retina has remained attached for a minimum of six months from the last retinal procedure. He notes that 10 (43%) of 23 eyes in our report had follow-up of only one or two months. Tables 1-5 list the results in 17 (74%) eyes on which we have follow-up of at least six months (mean = 9.3 months). Of the six patients on whom we do not have at least six months follow-up, two have left the United States, and the whereabouts of four are unknown. The follow-up of these six patients ranged from one to four months after surgery.

Table 1 Surgical Indications

Indication*	Eyes with > 6 months follow-up	All operated eyes
TRD + VH	7/17 (41.2%)	11/23 (47.8%)
TRD	3/17 (17.6%)	4/23 (17.4%)
TRRD	2/17 (11.8%)	2/23 (8.7%)
TRD + FS	1/17 (5.9%)	2/23 (8.7%)
TRD + NVG	1/17 (5.9%)	1/23 (4.3%)
TRD + NVG + VH	1/17 (5.9%)	1/23 (4.3%)
TRD + VH + FS	1/17 (5.9%)	1/23 (4.3%)
TRD + FS + NVG	1/17 (5.9%)	1/23 (4.3%)

*TRD, traction retinal detachment; VH, vitreous hemorrhage; TRRD, traction-rhegmatogenous retinal detachment; 
FS, fibrinoid syndrome; NVG, neovascular glaucoma.

The composition of the subgroup with at least six months follow-up strongly resembles the composition of the entire group of 23 eyes.[1] For example, the indications for surgery are almost identical in their representation (Table 1). The average complexity score in this subgroup was 4.5, which is the same as that of the entire cohort. Additional details regarding this cohort of 17 eyes are as follows. In eight (47%) eyes, silicone oil infusion (SOI) was part of the initial operation during primary vitrectomy. In nine (53%) eyes, SOI was done after previously failed vitrectomy. Preoperatively, 11 (65%) eyes were phakic, and six (35%) were pseudophakic. Eight (73%) of 11 phakic eyes underwent lensectomy, and five (29%) of these were left aphakic. Four (67%) of the pseudophakic eyes underwent intraocular lens removal and were left aphakic. Different intraocular procedures were performed at the time of surgery. Scleral buckles were placed in two (12%) eyes. Epiciliary dissection was done in two (12%) eyes, one of which had preoperative hypotony. Membrane peeling was done in all eyes. A Baerveldt valve was placed in one (6%) eye, and a peripheral relaxing retinectomy was necessary in one (6%) eye. Silicone oil (SO) was removed in two (12%) eyes at five and eights months after surgery, respectively, without complication. SO was removed due to migration into the anterior segment with corneal touch in one eye and due to cataract progression in the second eye. As one might expect, in all these aspects, this cohort of 17 eyes with at least six months follow-up resembles the overall cohort of 23 eyes on which we reported.[1]

Table 2 Anatomic results of vitrectomy, membrane peeling, and silicone oil infusion.

Cohort	Retinal reattachment rate
Eyes with > 6 months follow-up
Overall	12/17 (71%)
PPV +SOI as primary intervention	7/8 (88%)
PPV + SOI for failed previous vitrectomy	5/9 (56%)
All Eyes
Overall	17/23 (74%)
PPV +SOI as primary intervention	9/11 (75%)
PPV + SOI for failed previous vitrectomy	8/11 (73%)

The anatomic results of surgery among eyes with at least six months follow-up are quite similar to those of the entire cohort although a lower reattachment rate occurred among eyes undergoing surgery after having failed previous vitrectomy (Table 2). After a single operation employing pars plana vitrectomy with membrane peeling (PPV) and SOI, the retina was fully attached in 12 (71%) of 17 eyes with at least six months follow-up. In the eight eyes that were operated for the first time with primary SOI, the retina was attached in seven (88%) eyes. In the nine eyes that underwent SOI after previously failed vitrectomy, the retina was attached in five (56%). The difference in the reattachment rate between these two groups was not statistically significant (Fisher’s exact test, p = 0.18) as was the case in the cohort of 23 eyes.[1]

Among eyes with at least six months follow-up, three of the five eyes with recurrent detachment underwent repeat PPV and SOI. The retina was reattached in two (40%) eyes. The third eye developed hypotony despite complete retinal attachment five months after SOI and underwent repeat surgery with additional SOI and epiciliary dissection but became phthisical. The remaining two patients declined additional surgery. Therefore, with multiple operations employing PPV and SOI, the final anatomic success rate at last follow-up was 14/17 (82%), which is close the final anatomic success rate (20/23 (87%)) we reported previously.[1]

Table 3 Visual outcome.

Visual outcome	Eyes with > 6 months follow-up	All Eyes
Improved	4/17 (23.5%)	10/23 (43.5%)
Unchanged	10/17 (58.9%)	3/23 (13.0%)
Worse	3/17 (17.6%)	10/23 (43.5%)

With longer follow-up, visual results were not as good as we reported initially (Table 3). Among eyes with at least six months follow-up (and as noted in our initial report[1]), two (9%) had no light perception at last follow-up (neither underwent scleral buckling). Both eyes had preoperative vision of light perception. Loss of light perception was due to retinal and optic nerve ischemia in each case. Preoperatively, the visual acuity of all patients ranged from light perception to less than counting fingers at one foot. Six months or more after surgery, three (18%) of 17 eyes had vision: less than or equal to 5/200. Among the entire cohort of 23 eyes, five (22%) had vision: less than or equal to 5/200.

Table 4 Regression of rubeosis iridis and neovascular glaucoma.

	Eyes with > 6 months follow-up	All Eyes
Rubeosis iridis regression	5/7 (71%)	5/7 (71%)
Neovascular glaucoma regression	2/3 (67%)	2/3 (67%)

Rubeosis iridis was present preoperatively in seven (30%) of 23 eyes, three of which had neovascular glaucoma (NVG). The rate of NVG regression among eyes with at least six months follow-up was the same as in the entire cohort (Table 4). The rate of rubeosis iridis regression was greater than we reported initially (i.e. 3/7 (43%)) due to continued regression of rubeosis during the longer period of follow-up (Table 4). Of the three eyes with NVG, one eye underwent Baerveldt valve placement with normalization of the postoperative intraocular pressure. Five months after surgery, however, the eye developed hypotony. Subsequently, the valve was removed, and the eye underwent additional SOI. As noted above, however, the eye became phthisical. As reported initially, the NVG regressed after surgery in the second eye, and the third eye had NVG regression but no light perception postoperatively due to ischemia. As noted in our initial report, only one eye developed de novo rubeosis iridis.[1] Among the 17 eyes with at least six months follow-up, six (35%) had intraocular pressure: less than or equal to 5 mmHg. In our initial report, five (23%) of 23 eyes had intraocular pressure: less than or equal to 5 mmHg.[1] Three of these six eyes had persistent retinal detachment, which we presume to be the cause of the hypotony. Among eyes with at least six months follow-up, the remaining 11 (65%) had intraocular pressure ranging from 6-48 mm Hg, with four patients taking anti-glaucoma medications.

Table 5 Prevalence of complications.

Complication	Eyes with > 6 months follow-up	All Eyes
Intraoperative retinal break	3/17 (18%)	4/23 (17%)
Silicone oil in anterior chamber	3/17 (18%)	1/23 (4%)
Cataract	3/17 (18%)	1/23 (4%)
Hypotony with attached retina	3/17 (18%)	2/23 (9%)
Fibrinoid reaction	1/17 (6%)	1/23 (4%)
New onset rubeosis iridis	1/17 (6%)	1/23 (4%)
Keratopathy	1/17 (6%)	2/23 (9%)

Intraoperative complications were not different among eyes with at least six months follow-up vs. the entire cohort of 23 eyes (Table 5). Postoperative complications differed in that there was an increased prevalence of cataract, hypotony, and silicone oil in the anterior chamber over time, which is not surprising.

Table 6 Condition of fellow eye.

Condition	Prevalence
Proliferative diabetic retinopathy	21/21 (100%)
Status post panretinal photocoagulation	21/21 (100%)
Status post vitrectomy	10/21 (48%)
Clinically significant macular edema	3/21 (14%)
Traction retinal detachment	3/21 (14%)
Rubeosis iridis	2/21 (10%)
Neovascular glaucoma	2/21 (10%)
Central retinal vein occlusion	1/21 (5%)
Macular pucker	1/21 (5%)
Primary open angle glaucoma	1/21 (5%)
Visual acuity
20/20-20/50	6/21 (29%)
20/60-20/100	7/21 (33%)
20/200-20/400	2/21 (10%)
Worse than 20/400	6/21 (29%)

Professor McLeod suggests that in addition to using data from eyes with at least six months follow-up, one should use the status of the fellow eye to judge surgical success. At the time of surgery, all fellow eyes had proliferative diabetic retinopathy (in addition to other vision threatening conditions) and had undergone full panretinal photocoagulation (Table 6). Among patients with traction retinal detachment in the fellow eye, two underwent fellow eye surgery, and one patient refused surgery (Table 6). Six (29%) of 21 fellow eyes were pseudophakic. Among the six fellow eyes with visual acuity Professor McLeod points out that only seven (30%) of 23 eyes in our series had retinal breaks (either pre-existing or intraoperative), and he asks what the rationale for silicone oil use was in the remaining 16 eyes. It is not our habit to use silicone oil routinely for re-operation of eyes with proliferative diabetic retinopathy. We usually manage such cases with membrane dissection, extensive laser photocoagulation, and gas tamponade, when indicated.[3] The average case complexity score in a series of more than 150 eyes with severe diabetic retinopathy on which we operated using modern surgical techniques was 3.7.[3] In this series, the average complexity score was 4.5, which was significantly greater.[1] We felt that silicone oil was needed in all these cases due to the severe nature of the proliferative diabetic retinopathy as well as specific features of the case.

Specifically, rubeosis iridis was present in seven (30%) eyes, despite previous application of substantial panretinal photocoagulation, and silicone oil was used to compartmentalize the eye and inhibit progression of rubeosis iridis (Table 4). Among six (26%) eyes with retinal breaks, silicone oil tamponade was used due to the extensive nature of the retinal breaks. Normally, we use intraocular gas for this purpose.[3] Five (22%) eyes had traction retinal detachment and anterior hyaloidal fibrovascular proliferation, and silicone oil was used to help maintain a more normal intraocular pressure and forestall the development of phthisis. In two (9%) eyes with the fibrinoid syndrome, silicone oil was used to prevent the development of transvitreal fibrous sheets, which we feared might foster recurrent retinal detachment (Tables 1 and 2). Two eyes (9%) had no light perception in the fellow eye, and silicone oil was used to provide more rapid visual rehabilitation (Table 6). One eye (4%) had recurrent vitreous hemorrhage, and silicone oil was used to help maintain media clarity postoperatively.

Professor McLeod observes that silicone oil continues to be used during diabetic vitrectomy despite the attendant anterior and posterior segment complications. He suggests that wide-angle viewing systems, heavy liquids, endolaser, and long-acting gases enable one to manage posterior retinal breaks and retinectomies. Professor McLeod asks, “What is the appropriate use of silicone oil in the diabetic eye in the modern era?” He suggests that in cases where breaks can be managed with gas tamponade and rubeosis can be reversed with retinal reattachment and laser photocoagulation, the use of silicone might be “gratuitous”. Professor McLeod suggests that appropriate uses of silicone might include patients with posturing difficulties or patients in whom there is a need for early visual rehabilitation.

We recognize that the conclusions of our study are limited because it is a nonrandomized retrospective study without a control group. Thus, we cannot identify the “virtuous” indications for the use of silicone oil in the setting of severe proliferative diabetic retinopathy based on these data. Without the use of silicone oil, for example, rubeosis iridis might have regressed, and the retina might have remained attached in eyes exhibiting the fibrinoid syndrome. Nonetheless, the data from our study are consistent with the notion that silicone oil is an acceptable and useful tool in the management of eyes with severe complications of proliferative diabetic retinopathy. Our experience suggests, but does not prove, that silicone oil tamponade improves the prognosis in some otherwise unsalvageable cases. Among 11 (48%) of 23 eyes in this series, silicone oil was used initially at the time of repeat vitreous surgery. Short-term retinal reattachment was achieved in eight (73%) of these eyes. Among these 11 eyes, nine had follow-up of at least six months, and retinal reattachment was maintained in five (56%) of these nine with a single operation. Overall, 10 eyes failed initial PPV with or without SOI, underwent repeat PPV and SOI, and had at least six months follow-up. Retinal reattachment without phthisis was achieved in seven (70%) of these eyes. Since the initial vitrectomies employed modern surgical techniques, these results indicate that even in the modern surgical era, use of silicone oil can improve anatomic (and functional) outcome in selected cases.

References

(1) Castellarin A, Grigorian R, Bhagat N, Del Priore L, Seery C, Zarbin MA. Vitrectomy with silicone oil infusion in severe diabetic retinopathy. Br J Ophthalmol 2003; 87(3):318-21.

(2) McLeod D. Silicone oil in diabetic vitrectomy [electronic response to Castellarin A et al. Vitrectomy with silicone oil infusion in severe diabetic retinopathy] bjphthalmol.com 2003 http://bjo.bmjjournals.com/cgi/eletters/87/3/318#130

(3) Grigorian R, Castellarin A, Del Priore L, Fegan R, Zarbin MA. Epiretinal membrane removal in diabetic eyes: comparison of viscodissection with conventional methods of membrane peeling. Br J Ophthalmol 2003; in press.

Dear Editor

Castellarin and colleagues [1] recount their recent experience of infusing silicone oil in a small series of patients with advanced diabetic eye disease, either during primary vitrectomy (12 eyes) or after earlier surgery had failed (11 eyes). They compare their results with previous reports and conclude that silicone oil remains a useful adjunct in diabetic vitrectomy. However, their conclusions and historical comparisons are open to question.

Silicone oil was first used in primary diabetic vitrectomy in an era (1979-84) before the introduction of endolaser and the Landers' double concave lens for phakic fluid: air exchange.[2-4] Dealing with large or multiple posteriorly-located breaks (whether pre-existing or iatrogenic) was problematic, and direct fluid:silicone oil exchange (by virtue of the optical advantages of oil over air in the phakic eye) provided a surgical escape route, obviating the need for lensectomy. Furthermore, the clarity of the media immediately postoperatively facilitated the slit-lamp delivery of focal laser in order to seal retinal breaks that had been closed by the internal tamponade, and in addition the application of scatter laser to re-attached, untreated, ischaemic retina that had undergone deturgescence, in part through the 'waterproofing' effect of silicone oil.[3,4] All being well, the silicone oil could then be removed shortly thereafter, and some eyes that would undoubtedly have been lost were saved by the intervention of silicone oil in this way. Often, however, there were considerable associated problems, not least the rapid development of reparative epiretinal fibrosis whereby the retina re-detached under tangential traction and/or from re-opening of retinal breaks.[2-6] Sometimes huge areas of retinal disintegration eventually developed.[5,7] The fibroglial epiretinal proliferation appeared (both clinically and pathologically) to be particularly induced by clotted blood trapped between the silicone oil and the retinal surface or, ironically, by fibrin released as a result of the extensive scatter laser that was often needed to prevent highly vascularised membranes from re-proliferating behind the silicone oil.[4,5,8,9]

It was hoped that the so-called 'compartmentalisation' of the eye by silicone oil (to which the retro-silicone oil neovascularisation was attributed) might in turn result in prevention or reversal of rubeosis iridis through its putative barrier effect against anterior diffusion of angiogenic substances derived from the ischaemic retina.[3-8] Paradoxically, eyes with successful retinal reattachment (albeit with unabated ischaemia) often underwent rapid development or progression of iris neovascularisation,[3,8] while those with failed surgery from postoperative rhegmatogenous recurrence of retinal detachment (and therefore eyes with an exaggerated angiogenic drive) had evidence of protection from rubeotic phthisis, at least in the short term.[3] Perhaps naively it was postulated that rhegmatogenous confinement of the re-detachment by intravitreal silicone oil (and the consequent 100% oil filling of the shrinking vitreous cavity) might allow an effective obstruction to anterior molecular diffusion to be established in these failed cases.[3] Others had planned from the outset to employ silicone oil in their surgical protocol, not least for those diabetic eyes wherein earlier vitrectomy had been unsuccessful as a consequence of retinal re-detachment [4,10,11] or recurrent vitreous cavity haemorrhages.[4] However, whether used during primary diabetic vitrectomy or secondarily, whether unpremeditated or planned, and whether infused by direct fluid:oil exchange or sequential fluid:air and air:oil exchanges, the possibility of silicone oil limiting rubeosis and maintaining macular attachment despite peripheral retinal re-detachment was always welcome, even if surgical 'success' (that is, retinal attachment through 360 degrees) had strictly been denied.[2-4,8,12]

Nowadays, posterior retinal breaks and retinectomies can generally be managed successfully by employing wide-angle viewing systems, heavy liquids, endolaser, and long-acting gases. However, silicone oil continues to be infused during diabetic vitrectomy despite the attendant posterior segment and anterior segment complications that have only been partially mitigated by the improved quality of the silicone oil. The important question that thus arises is: what is the appropriate use of silicone oil in the diabetic eye in the modern era? Where retinal breaks might be closed just as readily using gas tamponade, or where rubeosis iridis might be reversed or prevented by retinal reattachment and/or a sufficiency of scatter laser photocoagulation, the use of silicone oil might be described fairly as 'gratuitous'. Exceptions might include anticipated posturing difficulties [3] or the need for early visual rehabilitation in one-eyed patients.[4] However, recent reports documenting the use of silicone oil in diabetic vitrectomy have failed to provide clear criteria or explanations regarding case selection.[1,12,13] Only 7 of the 23 eyes in Castellarin and colleagues' series, for example, had retinal breaks (2 pre-existing, 4 iatrogenic and 1 retinectomy), so the need for prolonged internal break tamponade was presumably not an issue in the majority of their eyes. More information is needed on the rationale for silicone oil infusion (not just the overall indications for surgery) in the remaining eyes in order to enable the potential benefits of this surgical adjunct to be assessed at this time. Furthemore, surgical success can really only be judged after a minimum of six months from the last vitreoretinal procedure,[3,4,8,10-12] and that judgement should preferably include consideration of whether the silicone oil has been removed and the status of the fellow eye.[14] The fact that 10 of the 23 eyes in Castellarins series were followed for only one or two months was thus a further serious limitation of their study.[1]

Infusion of silicone oil can be a most beguiling option during the closed microsurgical management of the stricken diabetic eye but, as mentioned, complications are prone to accumulate with time. Distinguishing the gratuitous from the virtuous use of silicone oil can be problematic, and equally it may be difficult to define the line between a surgeon's infusing silicone oil in anticipation of eventual surgical failure and such infusion representing his/her unstated admission that surgical failure has occurred already. All these issues need to born in mind when making historical comparisons between case series and in defining the place in history for silicone oil in diabetic vitrectomy.

References

(1) Castellarin A, Grigorian R, Bhagat N, et al. Vitrectomy with silicone oil infusion in severe diabetic retinopathy. Br J Ophthalmol 2003; 87: 318-21.

(2) Lean JS, Leaver PK, Cooling RJ, McLeod D. Management of complex retinal detachments by vitrectomy and fluid/silicone exchange. Trans Ophthalmol Soc UK 1982;102: 203-5.

(3) McLeod D. Silicone-oil injection during closed microsurgery for diabetic traction retinal detachment. Graefe's Arch Clin Exp Ophthalmol 1986; 224:55-9.

(4) Lucke KH, Foerster MH, Laqua H. Long-term results of vitrectomy and silicone oil in 500 cases of complicated retinal detachment. Am J Ophthalmol 1987; 104: 624-33.

(5) Barry PJ, Hiscott PS, Grierson I, et al. Reparative epiretinal fibrosis after diabetic vitrectomy. Trans Ophthal Soc UK 1985; 104: 285- 96.

(6) Charles S. Vitreous surgery, 2nd edition Baltimore: Williams and Wilkins, 1987. Pp 115-31.

(7) Wilson-Holt N, Gregor Z. Spontaneous relieving retinotomies in diabetic silicone filled eyes. Eye 1992; 6: 461-4.

(8) Yeo JH, Glaser BM, Michels RG. Silicone oil in the treatment of complicated retinal detachments. Ophthalmology 1987; 94: 1109-113.

(9) McLeod D, James CR. Viscodelamination at the vitreoretinal juncture in severe diabetic eye disease. Br J Ophthalmol 1988; 72: 413-9.

(10) Rinkoff JS, de Juan E, McCuen BW II. Silicone oil for retinal detachment with advanced proliferative vitreoretinopathy following failed vitrectomy for proliferative diabetic retinopathy. Am J Ophthalmol 1986; 101: 181-6.

(11) Brourman ND, Blumenkranz MS, Cox MS, Trese MT. Silicone oil for the treatment of severe proliferative diabetic retinopathy. Ophthalmology 1989; 96 : 759-64.

(12) Azen SP, Scott IU, Flynn HW Jnr, et al. Silicone oil in the repair of complex retinal detachments. A prospective observational multicenter study. Ophthalmology 1998;105: 1587-97.

(13) Scott IU, Flynn HW Jnr, Lai M-Y, Chang S, Azen SP. First operation anatomic success and other predictors of postoperative vision after complex retinal detachment repair with vitrectomy and silicone oil tamponade. Am J Ophthalmol 2000; 130: 745-50.

(14) McLeod D. Microsurgical management of neovascularisation secondary to posterior segment ischaemia. Eye 1991; 5: 252-9.